MiniMax H3 video generation API and CLI for AI agents
Open-source AI video generator for text, image, and audio to video creation
Stage
Not given
Not given
Founded
Not given
Not given
Based in
SG
Not given
Pricing model
Paid
Freemium
Pricing
Night (22:00–10:00 SGT): $0.009 per second or 6 credits per second. Day (10:00–22:00 SGT): $0.0135 per second or 9 credits per second. 1080P costs double. Credit packages: 10,000 credits for $15, 50,000 credits for $75, 100,000 credits for $150, or custom amount from $15 up to $1,500. 1 credit = $0.
Not given
Key features
Text to video generation from text prompts
First frame to video with optional final frame selection
Free shot video framing from character or photos
Lip sync for photos with voice or singing
Video length from 4 to 60 seconds
Multiple aspect ratios: vertical, horizontal, and square
Resolution options at 768P or 1080P
CLI for AI agents like Claude Code and Codex
HTTP API for programmatic access
Web studio for direct video creation
Text-to-video generation from natural language prompts
Image-to-video transformation with dynamic motion and animation
Audio-to-video synchronization with tight lip-sync
Native portrait video at 1080x1920 resolution
22 billion parameter Diffusion Transformer engine
4x-expanded text connector for accurate prompt interpretation
Face and character preservation across frames
Multi-modal generation pipeline with depth conditioning
Rebuilt VAE for sharper output and better texture preservation
Video output up to 1080p in multiple aspect ratios
What makes it different
Up to 80% cheaper than official MiniMax API
Pay only for seconds of video generated, not per request
Seamless integration with AI agents with one sentence setup
No credit card required to start with 500 free credits
Credits never expire
18x faster throughput than competing open-source models on H100 GPUs
Cloud-based rendering with no local GPU or installation required
Free credits for new accounts with no signup friction
Full commercial rights with no watermarks or royalty fees
Open-source model weights available on Hugging Face for local use